# Guide

# Introduction

InTween is a toolkit for creating interactive animations by tweening (opens new window). It performs similar functionality to other libraries like TWEENJS (opens new window), animejs (opens new window), and tween.js (opens new window).

However InTween excels at creating interactive, and interrupted tweens. A great example of where you would do this is in creating interactive videos, like this one (opens new window).

The fastest way to get started is to install InTween and follow the TLDR instructions.

Many seasoned tweeners will be familiar with a tween.js (opens new window) style of animating things, so let's see how the self-contained example in their readme would be implemented with InTween:

The Tween.js Code
const box = document.createElement('div')
box.style.setProperty('background-color', '#008800')
box.style.setProperty('width', '100px')
box.style.setProperty('height', '100px')

// Setup the animation loop.
function animate(time) {

const coords = {x: 0, y: 0} // Start at (0, 0)
const tween = new TWEEN.Tween(coords) // Create a new tween that modifies 'coords'.
  .to({x: 300, y: 200}, 1000) // Move to (300, 200) in 1 second.
  .easing(TWEEN.Easing.Quadratic.Out) // Use an easing function to make the animation smooth.
  .onUpdate(() => {
    // Called after tween.js updates 'coords'.
    // Move 'box' to the position described by 'coords' with a CSS translation.
    box.style.setProperty('transform', `translate(${coords.x}px, ${coords.y}px)`)
  .start() // Start the tween immediately.

Here's the InTween way to do it:

const box = document.createElement('div')
box.style.setProperty('background-color', '#008800')
box.style.setProperty('width', '100px')
box.style.setProperty('height', '100px')

const tween = new InTween.Tween({ x: 0, y: 0 })
  .by('1s', { x: 300, y: 200 }, 'quadOut') // Move to (300, 200) in 1 second.
  // ...using an easing function to make the animation smooth.

// Start the tween immediately.
  .subscribe((state) => {
    // Move 'box' to the position described by 'state' with a CSS translation.
    box.style.setProperty('transform', `translate(${state.x}px, ${state.y}px)`)

# Installation

# ES6 / Webpack / npm

Like most other libraries these days, just install it with yarn/npm and bundle it with webpack (opens new window) or rollup (opens new window).

Now you can import it and go!

import { Tween, Meddle, Player } from 'intween'

# Old-School Script Tags

Feel free to include the library manually with a script tag. All functionality will be kept inside a global InTween variable.

<script src="path/to/intween.min.js"></script>
const { Tween, Meddle, Player } = InTween // window.InTween


You can also get InTween from a CDN:

<script src="https://unpkg.com/intween/dist/intween.min.js"></script>

# TLDR; (Too Long Didn't Read)

The sections after this one will go into MUCH more detail. But to quickly get something up and running, this is what you need:

  • A Tween
  • A Meddle
  • A Player
  • spreadAssign()
  • animationThrottle()
import { Tween, Meddle, Player, spreadAssign, animationThrottle } from 'intween'

Here's an example of creating a tween that changes a position from [0, 0] to [1, 1] between 1s and 2s with quadratic easing:

const tween = new Tween({
  position: [0, 0]

  position: [1, 1]
}, {
  startTime: '1s',
  endTime: '2s',
  easing: 'quadInOut'

To create a meddle to be able to override the position (eg: by user input) with backIn easing, do this:

const meddle = new Meddle(tween).easing('backIn')

// connect it to some interaction event
window.addEventListener('click', e => {
  const position = [e.clientX, e.clientY]
  meddle.set({ position })

Now you need to create a player to "play" the tween and the meddle and subscribe to the output.

const player = new Player(tween.duration)

  , animationThrottle()
).subscribe(state => {
  // do stuff with state.position

player.play() // go!

# What is a Tween, really?


A Tween (in InTween) can be used to transition properties between many states with easing functions.

Many coders will be familiar with using tweens in other libraries. The word "tween" is short for inbetween. Tweens are a way of moving through states inbetween a starting and ending state. A state could be anything; a boolean, a string, a position vector.

InTween formalizes this idea a bit. Think of a Tween as something that takes time as input, and gives state as output. They do not animate states all on their own. They simply calculate the inbetween state for a given time.


Tweens need to be paired with a time source to actually do animation.

time -> Tween -> state

So by specifying what you want the state to be, and when you want it to get there, a tween will calculate all the inbetween steps. This makes animation easy!


const tween = new Tween({ x: 0 }).by('5s', { x: 5 })
// NB: time is in ms
tween.at(0)    // => { x: 0 }
tween.at(1000) // => { x: 1 }
tween.at(2000) // => { x: 2 }
// ...
tween.at(5000) // => { x: 5 }

# Frames


A Frame is a target state that tweens transition between.

Some libraries let you chain their version of tweens into timelines to get a more complex multi-stage animation... This isn't necessary with InTween. Just keep specifying more frames and the tween will connect the dots.


Tweens can have many steps (aka "frames") to them. You'll almost never need more than one tween for one whole animation.

There are three main methods to specify frames:

  • Tween.by(): By a certain time, the state should be...
  • Tween.in(): In a certain time from the end of the last frame...
  • Tween.to(): Set the state, and manually specify the timing information.
tween.by('6s', { value: 6 }) // by t = 6s, value should be 6
tween.in('1s', { value: 7 }) // in 1s (t = 7s), value should be 7
  { value: 8 },
  { startTime: '7.5s', endTime: '8s' }
) // starting at 7.5s and ending at 8s transition the value from 7 to 8

# Durations

By default tweens use the full time between frames to transition (duration = "100%"). But it doesn't need to! You can specify different durations.

  { value: 8 },
  { startTime: '10s', duration: '0.5s' }
// ... same as ...
tween.by('8s', '0.5s', { value: 8 })

You can even specify durations as percentages.

tween.by('1s', { value: 1 })
// use 10% of the time (1s) to transition to this next frame
tween.by('11s', '10%', { value: 2 })

# Time Formats

Times can be specified in many formats. If specified as a primitive number, it will be interpreted to be in milliseconds (ms). If specified as a string, you can use the following formats:

"##:##" // minutes:seconds
"##:##:##" // hours:minutes:seconds
"#.#s" // # of seconds
"#.#m" // # of minutes
"#.#h" // # of hours
// durations can be expressed as percentages
"#.#%" // percentage of time between two frames


tween.in(2000, 1000, { value: 1 }) // in 2s (2000ms), transition duration 1s
tween.in('2s', '1s', { value: 1 }) // ... same
tween.in('10:10', '3:00', { value: 1 }) // in 10 minutes 10 seconds, duration 3 minutes
tween.in('10:10', '20%', { value: 1 }) // same with duration of 2 minutes 2 seconds (20%)

# Easing


Easing functions modify how a transition "feels". They change the way that one state transitions to another.

Animation would be pretty boring without easing. Easing is the quality of the transition between two states.

You can try out some easing functions here to see what they look like:


For more details, see the In Depth Documentation on State Transitions

The simplest easing function is linear, which is also the default:

// Behold, the linear easing function
const linear = n => n

But there are so many more possible ways to get from point A to point B. For a nice visual reference of standard easing functions, check out easings.net (opens new window).

InTween contains a set of easing functions adapted from the Phaser HTML5 game framework (opens new window).

The easiest way to use them is to just specify their name as a string for any InTween object methods.

For example:

tween.by('1s', { value: 1 }, 'bounceOut') // use bounce easing
tween.to({ value: 1 }, { endTime: '1s', easing: 'bounceOut' })

However you can also import and use them directly...

import { Easing } from 'intween'
tween.by('1s', { value: 1 }, Easing.bounceOut) // use bounce easing

Some easing functions can be constructed to customize them more. For example, to take discrete steps, you can use the makeSteps(n) easing factory.

import { Easing } from 'intween'
tween.by('1s', { value: 1 }, Easing.makeSteps(5)) // take 5 discrete steps


It's also possible to combine easing functions using pipe or Util.combineEasing.

You can also combine easing functions for more complex behaviour. Although, success can be a bit hit or miss in terms of smoothness. To combine easing functions using them one after another, you can reference them by name:

tween.by('1s', { value: 1 }, 'quadIn + backOut')

or use Util.combineEasing:

import { Util, Easing } from 'intween'
const combined = Util.combineEasing(
tween.by('1s', { value: 1 }, combined)

Sometimes you can get better results by using pipe(), which is the same as doing k => Easing.bounceOut(Easing.quadIn(k)).

import { pipe, Easing } from 'intween'
const fall = pipe(
tween.by('1s', { value: 1 }, fall)

# Time Sources


Time sources like Player and animationFrames provide the time input used by Tween and Meddle.

In the InTween library, Tween objects by themselves don't do anything. They are just a way of mapping time to state. If you want an inbetween state for a given time, you can call Tween.at(t) for a given time to get the state.

If you really wanted you could do something like this:

// a cumbersome way of animating...
const tween = new Tween({ x: 0 })
  .by('2s', { x: 2 }, 'quadInOut')
  .by('4s', { x: 0 }, 'quadInOut')

const startTime = window.performance.now()
const onAnimationFrame = () => {
  const time = window.performance.now() - startTime
  const state = tween.at(time)
  // update display using state
  // ...

onAnimationFrame() // start

The above code sets up a tween, and gets the current state every frame using requestAnimationFrame. If you want, you could use InTween like this. But there is a much nicer way...

# Observables

Tweens are meant to be used as Operators for time Observables (opens new window), which we'll elaborate on here.


InTween is built on top of the concept of Observables (opens new window) and works very well with RXJS (opens new window).

An Observable is very similar to a data stream. Most of the core Observables in InTween are time-based. So you can subscribe to them to react to changes in time. The most basic time Observable is created by calling animationFrames().

// Every animation frame, this prints out the
// current time since subscribe was called
const subscription = animationFrames().subscribe(time => {

// later... we can cleanup
subscription.unsubscribe() // stop


Be sure to clean up your subscriptions once you're done by calling the subscription's .unsubscribe() method.

So animationFrames() creates a time Observable that gives us the current time every animation frame. If we want the tween state, we just need to convert that time into state using the tween. We mentioned earlier that a Tween is just an Operator. What this means is that we can pipe the time through the tween operator to get the state every animation frame, like so:

import { animationFrames } from 'intween'
  .subscribe(state => {
    // update display using state

Wow! So much nicer. And also, much more versatile. But that's not all...

# Player

Sometimes all you need is a quick animation that plays all the way through... but other times, you want to be able to pause, seek, and replay your animation.

For more fine-grained control over the timing we can use a Player.

import { Player } from 'intween'
// create a player to match the duration of our tween
const player = new Player(tween.duration)

const subscription = player.pipe(tween)
  .subscribe(state => {
    // update display using state

// elsewhere...
player.seek(2000) // seek to 2s
player.playTo(3000) // play until 3s
player.on('end', () => {
  player.seek(0) // go back to start once ended
// cleanup when finished

Here's a little demo showing how you could build your own UI for a player and attach the player instance to a play/pause button, a scrubber, and a time display.

# Data Types and Interpolators

InTween was built to animate anything. In addition to connecting the dots for numbers, InTween can handle all kinds of primitive types; Number, Boolean, String. It can also handle objects with numeric properties and arrays of any known type!

For example:

const tween = new Tween({
  number: 0,
  boolean: false,
  string: "Hello",
  array: {
    type: Array,
    value: [0, 0, 0]
  object: {
    type: Object,
    value: { x: 0, y: 0, z: 0 }
.by('1s', {
  number: 1,
  boolean: true,
  string: "Hello World!",
  array: [1, 1, 1],
  object: { x: 1, y: 1, z: 1 }

# Interpolators

You can read In Depth about Interpolators but the short version is this: Interpolators tell you how to get the inbetween values for a given type.

All primitive types have default interpolators associated with them. Most of the time, you won't need to worry about interpolators for primitive types.

The boolean type is a bit of an exception, of course, because there is no value between true and false. The boolean type uses the toggle interpolator.

If we want we can use the toggle interpolator with numbers by specifying it when we create a tween.

The following example creates a tween with a myNumber property that only changes to its final value once the 10 seconds is up.

new Tween({
  myNumber: {
    type: Number,
    value: 0,
    interpolator: 'toggle'
.by('10s', { myNumber: 11 })

Angular Interpolators

When dealing with angles, sometimes it can be useful to use either the degrees or radians interpolators to automatically transition the shortest distance around the circle.

Just as with easing functions, interpolators can be referenced by name, or accessed by importing them:

import { Interpolators } from 'intween'

And some interpolators can be created for more customizability:

import { Interpolators } from 'intween'
new Tween({
  clock: {
    type: Number,
    value: 0,
    interpolator: Interpolators.makeCyclic(12) // mod 12
  halfWay: {
    type: String,
    value: 'Hello',
    interpolator: Interpolators.makeToggle(0.5) // change half-way through

# Custom Types

Although animations mostly revolve around changing numbers, those numbers can be abstracted into concepts like vectors, complex numbers, even quaternions. InTween can handle these by defining custom interpolators.

But instead of specifying the interpolator on every state property you need it for, you can just define your own custom type using registerType().

Let's say you had an amazing 3D library (opens new window) and wanted to animate position vectors from that library. You could create a custom type like so:

import { registerType } from 'intween'
  type: 'Vector3'
  , default: new THREE.Vector3()
  , interpolator: (from, to, k) => {
    return from.clone().lerp( to, k )

const tween = new Tween({
  cubePos: { type: 'Vector3' }
.by('1s', { cubePos: new THREE.Vector3(1, 2, 3) })

# Interactive Tweens (Meddling)


A Meddle can be used to override a Tween state temporarily and gracefully return to the correct Tween state at a later time.

So far we have been creating non-interactive animations. The main strength of InTween is its ability to interactively tween things.

The main idea behind interactive tweening is this: A tween's state is overriden by user interaction (called meddling), and then seemlessly returned to the main tween timeline.

Here's a cartoon depiction of the process:


The way we achieve this is by using a Meddle object. A Meddle object needs awareness of the tween it will be "meddling" with so that it can peek ahead at the state it needs to merge back with in future.

const meddle = new Meddle(tween)

Like the Tween object, it also operates on time Observables... that is to say, you can pipe() a time Observable through it like this:

  .subscribe(state => {
    // use the meddled state

The default meddle state is empty ({}), so this alone wouldn't do much. However if we call meddle.set() we can influence the current meddle state.

// on user interaction...
meddle.set({ x: userX })

Some things to note:

  1. Any call to .set() will trigger an override and the internal timer tracking when to rejoin the tween will reset.
  2. The property definitions of the meddle should match the tween it's meddling.
  3. The properties specified in a .set() call will be merged into the meddle's current state. So calling .set({ x: 1 }).set({ y: 2 }) is equivalent to .set({ x: 1, y: 2 }).

We can also influence how long we want to override the state, and how it should return to the tween state. We can do this ahead of time or when interaction happens.

const meddle = new Meddle(tween, {
  relaxDelay: '1s',
  relaxDuration: '2s',
  easing: 'bounceOut',
// -- and/or --

We can also prevent the meddle state from returning to the tween entirely by freezing the meddle where it is.

meddle.freeze() // stay like this
meddle.freeze(false) // release

# Merging Tween and Meddle States with spreadAssign()


The spreadAssign() function spreads the time input across several time Operators and then merges their resulting states together.

spreadAssign(A, B, C, ...): t -> A(t) ∪ B(t) ∪ C(t) ∪ ...

The Tween and Meddle states need to be used together and merged together. The way we can do this is by using a function called spreadAssign(). What it does is take the time input from a Time Source, feed it into multiple operators and use Object.assign() to merge the states together.

The standard way of combining a tween and a meddle would be something like this:

).subscribe(state => {
  // combined state

And if you need more meddles, just add them!

  /// ...

# Optimizing rendering with animationThrottle()


Placing animationThrottle() at the end of your state stream will prevent unnecessary reactions to state changes. (IE: it limits updates to the speed of window.requestAnimationFrame)

Normally, any changes to state will feed into the final subscribe() callback. If you have a bunch of meddle.set() calls, this may mean that the subscribe callback will get called faster than a normal 60 frames per second. This is unnecessarily wasteful on resources.

To correct this, we can use a special operator called animationThrottle() which simply removes the unnecessary extra updates. This makes it good practice to place it at the end of your pipe() right before you update the visuals of your animation.

).subscribe(state => {
  // combined state

# Smoothing User Interaction


By using smoothen() you can transition smoothly to target states on demand.

So far, the user interaction has overridden the state instantly. But often we'd like that to be a smooth transition from the tween state to the override state.

For example, if a user clicks on the screen we might want to move an object there smoothly. And if they click elsewhere during the transition, we'd like to respond to that smoothly too.

To do this we use smoothen(). Technically, this creates an operator that maps an Observable over Frames to animation states.

An example is called for...


A Subject is a quick way of creating observables. For more information on Subject, see the rxjs documentation on Subject (opens new window).

The above code would move a div element whenever you click on the page. And if you click while the div is currently moving, it will compensate for the new end position in a smooth way.


When using this with a Meddle you'll need to provide a state getter function to ensure that smoothen knows where to start from.

You can easily send the output of smoothen into a Meddle to make the meddling smooth! However, when using smoothen() to meddle with a tween, there's an extra step that needs to happen. Since a tween may change the state without smoothen() knowing about it, we need to provide a getter function so that it is aware of these changes.




// elsewhere...
let animationState
).subscribe(state => {
  animationState = state
  // update views...

/// smooth our meddles
    { duration: '1s', easing: 'quadOut' },
    () => animationState
).subscribe(state => meddle.set({ position: state.position }))

# Recipes and Extras

# Tweening Colors

When tweening colors, a nice way to do it is by using the amazing chromajs library (opens new window) and registering a new type for it.

# Using Audio Players (like Howler or SoundCloud) as Time Sources

As long as the audio player's api can tell you what time the audio is at (which is almost always the case) then you can create a timesource from it by using a combination of animationSync(), animationFrames() and map(). The animationSync() function ensures that even if there are small jumps in the precision of the reporting of the time, the animation will still come out smooth on the other side. The animationSync will also only report new time values which reduces costly rendering when paused.

Here's a funky demo that does this with Howler (opens new window).

If the audio player api provides a playback event (like SoundCloud does) then you can plug into that.

Here's how you might do use the soundcloud widget as a time source:

import { Subject spreadAssign, animationThrottle } from 'intween'

const scTimeSource = new Subject()

// somehow create a soundcloud widget, then...
widget.bind(SC.Widget.Events.PLAY_PROGRESS, e => {
  const time = e.currentPosition

  , animationThrottle()
).subscribe(state => {
  // ...

# Creating Slideshows

Here's a demo that shows a very easy way to create slideshows! The trick is to use player.playTo() so that it plays until the next slide. You can go all out and animate the slide content too, of course!

Last Updated: 9/14/2021, 6:54:46 PM